“All models are wrong, but some are useful” – George Box
Reality is incredibly complex. To handle it and reduce the complexity, we create models that zoom in on aspects of how the world works. They are abstractions that simplify reality and make it easier to understand the world and avoid overwhelm. How good an abstraction is, is always measured by its utility.
Imagine if when you wanted to buy a car, you had to say, “Honey, let’s go look at configurations of metals, polyester, and plastic.” That’d be highly impractical. The abstraction “car” summarizes all the usual components that make out a car and allow you to communicate the concept efficiently. Abstractions taken together form the mental models through which we see and interpret the world.
The danger with abstractions and models is that we forget what they leave out. Maybe a model leaves out the dynamics between its different parts. For example, a model of the inner organs of a human being does not show how these organs interrelate to each other. Sometimes, the world changes in a way so that the model stops being true. If we take models as gospel, we’ll act under wrong assumptions and make costly mistakes.
Forgetting that the model is not reality might lead us to be falsely confident about our understanding and our derived actions. While limiting complexity and simplifying reality at times is useful and necessary, we shouldn’t forget the risk this introduces in other contexts.
Maps are not the territory
The best example for this are maps. A map is an abstraction of the territory. It’s depicting certain aspects of reality, e.g. altitude levels, location of rivers and cities, but is also neglecting others. Maybe the map doesn’t include the depth of rivers and we consider them as impassable obstacles even though we could easily wade through them. The map of the London Underground is useful for passengers navigating how to get across the city. But it’s insufficient for architects to plan any type of construction.
The simplification is necessary for the map to be useful. If it would be comprehensive, the complexity would again overwhelm us. But while a map without the depth of rivers is useful in a variety of situations, it isn’t the best tool to find a way across rivers.
Renting an apartment
When I rented my first apartment, I found a listing of a place with floor plans and photographs of how it looks like. I knew of course that the photographs would only show the place in the best light and that the floor plans wouldn’t give me a feel for the size of the place. I needed to visit and check for myself.
After the visit I’ve expanded my model of the place . I knew how the space felt. I checked the windows towards the street for its soundproofing qualities. I looked at the general condition of the place. In the end, I took it, based on the online listing and a half hour visit.
Of course, even that updated model had its limitations. I later realized that one of the walls to a neighboring apartment was out of plywood. My neighbor and I got intimately accustomed to every part of our lives over the next few months. I since further improved my model and vowed to check every single wall of potential future apartments before making a decision.
Reality is the ultimate update
When using a model, its limitations should be top of mind. Using it in a context where it leaves out important parts of reality introduces considerable risk. When a model and reality clashes reality always wins. In other words, when a model predicts things to go differently than they are, it’s time to update the model.
While models are a representation of reality, they can also influence it. We make decisions based on our models, which in turn, creates our reality to a large part. People that believe their life to be determined by luck will act with less agency and hence, their lives will indeed be more determined by external circumstances. The same goes for groups and organizations as well. The ideologies of a society will determine the direction of it. When everyone believes the world to be flat, nobody will try to sail too far out into the ocean. Sometimes, for outcomes to change, we have to question, update, and ultimately change our models.
Consider the origin
Where does the model come from? Who developed it and in what context? Answering these questions will give you great insights into where the model is useful and what biases and limitations it includes. What incentives might the modeler have had, what was their goal in creating the model? Since, as we know, models can be a self-fulfilling prophecy, we must be aware of their ideological uses which move us away from accurately understanding reality.
No limitations. Abstractions and models never reflects reality fully. Ignoring this will inevitably lead to false confidence and faulty decision-making.
The portrait. The artist’s own perception enters the painting without any conscious thought. Some aspects of reality are heightened, some are omitted. It all depends on the context in which it was painted in, the painter, the lighting, the mood of the sitter, the purpose of the painting… The portrait will never reflect the reality fully, but it might fulfill the intended purpose of communicating the status of a nobleman, the memory of a deceased, the cult around a leader.